Broken quotes and mismatched rows.
Fix the earliest parsing problem before deciding a record is missing fields. One absent quote can change how many later lines are read.
A comma inside a value needs quoting.
In a comma-delimited file, 42,Smith, Ana,active has four fields. If the intended name is “Smith, Ana”, the record should be 42,"Smith, Ana",active. Only the original source or its owner can establish that intent.
Quotes inside quoted text are doubled.
The value She said "hello" is written as "She said ""hello""". A quote in the middle of an unquoted value is not the same as correctly escaped text. FieldWise points to the physical line where it found the problem.
A record can occupy more than one line.
A quoted note may contain a genuine line break. The next physical line is then part of the same logical record. The inspector labels that record by its starting line and shows its line range. It keeps those line breaks when converting.
A short record does not say where the missing field belongs.
Suppose the header is id,name,status and a record is 42,active. Is the name missing, or is “active” the name and the status missing? Padding at the end would choose one interpretation. The tool requires you to verify every short record before using that option. When in doubt, correct the source export.
Extra columns require inspection.
Unexpected delimiters or different record shapes can cause wider rows. The tool never truncates them automatically. Broken quotes and extra fields block downloads until you correct the text and inspect again. An empty trailing field such as 42,Ana, is kept; it is not automatically deleted.
What the structural check means.
A pass means the supported quoting and field-count checks found no errors under your selected options. It does not prove that your destination permits blank values, repeated headers or a particular business identifier. RFC 4180 documents a common CSV format; real applications use multiple dialects.